The 4 Phases of School AI Adoption and Policy

Elevate has offices across 4 countries, but we work directly with schools in over 20 countries, which gives us a genuinely global vantage point for viewing how schools are responding to the rise of generative AI. Across almost all of them there is one consistent trend: students are using AI. In our 2025 benchmarking survey of 3,000 high school students, we found that 75% were using AI on a weekly basis, with 25% using it daily. These figures are remarkably consistent across countries. The result for schools is that whether they like it or not, the genie is out of the bottle. AI is being used by their students, and that fact alone means schools need to consider how they respond.

Beyond the day to day reality of students using AI to assist with their work, schools face a second predicament: the transformation of the workplace and the economy by AI. Forecasts remain divided on the scale of that impact. In a May 2025 interview with Axios, Anthropic CEO Dario Amodei warned that AI could eliminate up to half of entry-level white-collar roles within the next one to five years. Others are more optimistic, arguing AI will create as many jobs as it displaces. Regardless of which view you take, the one thing almost every forecast agrees on is that the workplace will change, and change substantially. Schools are left having to make the philosophical decision as to what this means for them. To what degree, if any, should schools be equipping students for this changing landscape, and building their AI capability and literacy?

The question then is how schools are actually responding. We've found schools currently fit into one of four phases.

Phase 1: Haven't Started

From a recent survey of close to 100 high schools, we found 33% of schools sit in this category. Schools haven't started largely because developing a position on AI is genuinely difficult. Schools need to decide whether they're regulating AI, which generally means banning it outright, building a balanced approach that allows some use, or embedding AI literacy into the curriculum and classroom directly. Once a position is chosen, schools then need to work out what skills to upskill students and staff in, and how to actually deliver that capability building, which is its own significant undertaking. Stakeholder management adds another layer of difficulty. Few technologies are as polarising as AI. Staff are divided on whether it belongs in the classroom at all, and more surprisingly, so are students. Teenagers are usually the earliest and most enthusiastic adopters of new technology, but their response to AI has been noticeably different. Many students are wary of AI because of its environmental impact, its effect on creative industries, the risk of other students gaining an unfair advantage, the risk of being wrongly accused of cheating, and concern about not developing the skills they'll actually need in the future. Managing this spectrum of genuinely emotive positions takes careful stakeholder communication. Finally, once a position is settled, a policy needs to be developed to reflect and communicate it, and to manage the risks that come with introducing AI into a school. Even this basic list shows why it's easier to put off getting started, especially when every issue involves second and third order consequences and teachers don't have time to spare before implementation even begins.

Phase 2: Waiting for Guidance

The second reason schools haven't started is that they're waiting on guidance from a higher authority, whether that's a department of education, a diocese, or another governing body. There are two problems with this approach. Students are already using AI, and that use carries risk regardless of whether a school has taken a position. Banning AI at school doesn't remove this risk, since students will simply use it once they're off school grounds. As several schools have discovered, this can create significant reputational exposure, with a number of highly publicised cases making the news, ranging from academic misuse of AI through to AI being used to create deepfakes of students and staff, and for bullying. Schools that end up in the public eye over these issues consistently find that the time spent on crisis management and damage control is far greater, and considerably more stressful, than the investment required to start responding to AI proactively in the first place.

Phase 3: Responding to Issues

This is the phase most schools move through once an issue, generally academic misuse, forces their hand. In our survey of 100 schools, 62% were at this stage. For many, the first response is an outright ban, but schools quickly run into two problems. It doesn't solve the underlying issue, since students still use AI at home and misuse cases continue regardless. And proving misuse is genuinely difficult. Teachers often know intuitively when a student has used AI for part or all of a task, but proving it is another matter entirely. AI detectors are notoriously unreliable, with some tests showing false positive rates as high as 20%, and teachers are understandably hesitant to accuse a student of cheating without clear proof, meaning many cases simply go unaddressed.

Schools tend to evolve from an outright ban toward addressing the underlying causes of misuse. This starts with building students' AI skillset: understanding how the technology works, developing prompting skills, learning to verify and validate outputs, and knowing when AI use is and isn't appropriate. We cover this in full in our [PLACEHOLDER: link to AI Skills Inventory], and 77% of schools in our study named this upskilling as a current priority. The second and, we'd argue, more critical step is redesigning assessment so it's harder to bypass. As covered in last week's newsletter, this is close to the keystone task for schools at this stage. It doesn't just make misuse more difficult, it upskills teachers in AI literacy themselves, builds the processes needed to talk about AI's place in the classroom, and starts to shift school culture toward acceptance rather than resistance.

We believe this phase is the most transformative part of the whole process. It builds the initial skillset, gets staff interacting more with AI, and grows their confidence. Stakeholder divisiveness tends to decrease as a result, and organisational buy-in grows. This doesn't mean schools go all in and become AI-driven evangelists, but it does tend to lead naturally into Phase 4.

Phase 4: Integrating AI Literacy into the Classroom

Only 6% of schools in our survey were at this stage, but we believe schools that successfully redesign assessment inevitably progress here. This is where schools embrace the need to build genuine AI literacy and capability to prepare students for a workplace that's increasingly AI enabled. In this phase, schools broaden their skills inventories well beyond the foundational skills covered in Phase 3, extending into more practical, non-academic use cases students will need beyond school. They also start incorporating AI oriented tasks into daily teaching for two reasons: to deepen engagement, as staff confidence and openness to experimentation grows alongside their own AI use, and to build AI literacy in a genuinely interdisciplinary way, teaching course curriculum and AI literacy simultaneously rather than treating them as separate initiatives.

Moving Between the Phases

For most schools, Phase 4 is more aspiration than reality right now. Moving from Phase 1 to Phase 3, or successfully completing Phase 3, is time consuming, difficult, and can feel intimidating. Schools looking to make the move should start with three things.

First, define the outcome. Decide where you actually want to be as a school, and borrow from the process above rather than inventing your own from scratch. Don't try to jump straight from Phase 1 to Phase 4, since you're unlikely to have the stakeholder buy-in or the skills to support it. Instead, start small by identifying the specific problems you want to eliminate or reduce, especially academic misuse.

Second, determine where you actually are now. It's easy to build an aspirational skills wishlist that's disconnected from reality, so instead start by understanding how students and staff are currently using, and misusing, AI. Identify your biggest challenges first and build out the skills required to solve those specifically.

Third, understand your AI culture. Take the time to understand how your teachers, staff and parents actually feel and think about AI. This defines your real starting point for buy-in, and shapes how you approach your mission statement, communications and professional development from here.

Elevate can help with this process. We've spent the last 18 months surveying schools and students around the world to understand the challenges each is facing with AI and how best to respond. Want to see exactly where your school sits? Try a free module of our AI Culture Snapshot, benchmarked against the schools we work with, covering your school's AI strengths and weaknesses alongside what your initial and ongoing skills training should look like: Try free module.